The $4 Trillion Unrealized Stack Will Grow Before It Shrinks, and Engineered Liquidity Is Now Structural, Milken Panelists Argue

Read time: 4 minutes

What's New

Private equity's roughly $4 trillion of unrealized NAV will get larger before it gets smaller, so the liquidity tools built to work around it are permanent features of portfolio construction rather than a cyclical patch. Wil Warren, Partner and President of Lexington Partners, argues this on a panel at the Milken Institute Global Conference 2026, joined by Mubadala's Camilla Languille, Neuberger Berman's David Lyon, Kuvare's Dhiren Jhaveri, and Aquarian's Eric Murzyn. Distributions as a share of NAV ran at 25 to 28% a year between 2014 and 2020; last year it was 12%. Even a recovery to the high teens leaves the stack compounding. Allocators should plan pacing, underwriting, and governance for episodic liquidity, and treat single-asset CVs, NAV lending, and insurance balance sheets as the exit market.

Why It Matters

The industry narrative held that 2026 would restart the exit machine. Instead, a war in the Middle East chilled dealmaking and AI forced a rethink of software valuations. The conventional view is that IPO and M&A windows will reopen and the log jam clears. The panel's position is that valuation still has to adjust, that 2020 to 2021 vintages bought at peak multiples need longer holds, and that mid-market sponsors are stranded because the top 10 firms raise 65% of capital. Every panelist sells a liquidity solution, from secondaries to structured capital to insurance balance sheets, so the thesis is their market. Lyon adds the check: structure hedges valuation, it does not remove risk.

Big Picture Drivers

• Math sets the hold period: Distributions divided by NAV sit in the mid-teens, implying a hold near seven years. At 16 to 17 times adjusted EBITDA with 10 turns of equity, financial engineering cannot generate the return, so sponsors must strip, combine, or rebuild assets, which takes longer.

• Marks lag public markets: Public software has repriced about 30%, and Lyon has not yet seen an avalanche in private software marks. He treats "we will grow into our valuation" as an admission of mismarking.

• Capital formation is bifurcated: A $40 million EBITDA company that pushed price to $80 million during the pandemic needs an $800 million buyer at 16 times. Mega-caps do not want it and mid-market sponsors cannot raise for it. Sponsor-to-sponsor volume, the historical exit majority, has stalled.

• Insurance balance sheets need duration: Life insurers hold 30 to 40% of assets in private markets and must match 10 to 30-year liabilities. Kuvare invests about $6 billion of policyholder premiums a year and can wait through delayed draws that five-to-seven-year funds cannot.

• Sovereign direct capital sets its own price: Mubadala, at $385 billion AUM, has built sector and geographic teams to self-underwrite minority recaps and single-asset CVs, offering GPs DPI on assets they want to keep, at valuations Mubadala chooses.

• Cyclicality is the opportunity: Warren argues fundraising has slowed so much that a new entrant with primary capital can back the best GPs, buy portfolios across 15 vintage years, and set up attractive vintages, because the industry overfunds tops and underfunds bottoms.

By The Numbers

• ~$4 trillion: Unrealized private equity NAV, a figure Lyon notes everyone quotes and nobody can fully audit.

• 25 to 28% to 12%: Annual distributions as a share of NAV, 2014 to 2020 versus last year.

• 65%: Share of private equity fundraising captured by the top 10 firms.

• SOFR+540: Yield Kuvare and three co-investors structured on seasoned Blue Owl loans that had traded in the low 400s two months earlier; PIMCO financed the same vehicles a month later at +250.

• 20%: Share of exits last year that were GP-led secondaries, per Languille, a figure she expects to keep rising.

• 30% versus 13% and under 4%: Software exposure in private credit versus leveraged loans and high yield, the concentration that triggered redemption panic.

Key Trends to Watch

• Evergreen structures as a live experiment: Lexington runs about $30 billion of drawdown capital and takes in $1 to $2 billion a year of evergreen money into the same deals at the same pricing. Warren calls the redemption feature the untested part. Watch whether $25 billion retail PE vehicles become sellers when redemptions bite.

• Private credit secondaries at scale: The Blue Owl trade shows insurers buying seasoned loans near par when retail vehicles need cash. Watch whether the near-par bid persists as BDCs trade 20 points off and tenders at 20 to 30% discounts go unfilled.

• AI as the underwriting variable: Technology, professional services, and software make up roughly 45% of the US economy by Lyon's framing. Sifting the unrealized stack for durable business models is the hardest task in the market now.

• Regulatory attention on insurance private credit: Treasury and rating agency scrutiny of covenants, control rights, and rated feeders will shape which structures survive. Both insurers on stage say they welcome it.

Memorable Quotes

• "If the discount in our market were zero, volume would be infinite." Warren, quoting an old partner, on why price and volume in secondaries move together.

• "We're going to grow into our valuation. That means by definition it's mismarked." Lyon on the phrase he least likes to hear from sponsors.

• "First lien loans does not mean no risk. They mean first lien loans." Lyon on structure as a hedge rather than a panacea.

• "The unrealized stack of NAV that I keep referring to is going to get bigger before it gets smaller in any circumstance." Warren's closing forecast.

The Wrap

The thesis holds if DPI yields stay below the high teens through 2027, if GP-led secondaries keep growing past 20% of exits, and if insurers and sovereigns keep providing near-par liquidity for assets they can self-underwrite while the retail evergreen experiment survives its first redemption cycle. It fails if M&A and IPO windows reopen fast enough to return distributions to the mid-20s, which would make today's structured solutions a cyclical bridge rather than a permanent layer. It also fails if marks on 2020 to 2021 software assets finally reset and the bid-ask gap closes through price rather than structure. The 2026 to 2028 distribution yield, measured against a stack that Warren expects to keep compounding, is the number that settles it.